Diversification and spatial structuring in the mutualism between septica and its pollinating wasps in insular South East Asia Lillian Jennifer Rodriguez, Anthony Bain, Lien-Siang Chou, Lucie Conchou, Astrid Cruaud, Regielene Gonzales, Martine Hossaert-Mckey, Jean-Yves Rasplus, Hsy-Yu Tzeng, Finn Kjellberg

To cite this version:

Lillian Jennifer Rodriguez, Anthony Bain, Lien-Siang Chou, Lucie Conchou, Astrid Cruaud, et al.. Diversification and spatial structuring in the mutualism between Ficus septica and its pollinating wasps in insular South East Asia. BMC Evolutionary Biology, BioMed Central, 2017, 17, pp.207. ￿10.1186/s12862-017-1034-8￿. ￿hal-01589491￿

HAL Id: hal-01589491 https://hal.sorbonne-universite.fr/hal-01589491 Submitted on 18 Sep 2017

HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non, lished or not. The documents may come from émanant des établissements d’enseignement et de teaching and research institutions in France or recherche français ou étrangers, des laboratoires abroad, or from public or private research centers. publics ou privés.

Distributed under a Creative Commons Attribution| 4.0 International License Rodriguez et al. BMC Evolutionary Biology (2017) 17:207 DOI 10.1186/s12862-017-1034-8

RESEARCH ARTICLE Open Access Diversification and spatial structuring in the mutualism between Ficus septica and its pollinating wasps in insular South East Asia Lillian Jennifer Rodriguez1,2,3* , Anthony Bain4, Lien-Siang Chou4, Lucie Conchou2,5, Astrid Cruaud3, Regielene Gonzales1, Martine Hossaert-McKey2, Jean-Yves Rasplus3, Hsy-Yu Tzeng6 and Finn Kjellberg2

Abstract Background: Interspecific interactions have long been assumed to play an important role in diversification. Mutualistic interactions, such as nursery pollination mutualisms, have been proposed as good candidates for diversification through co-speciation because of their intricate nature. However, little is known about how speciation and diversification proceeds in emblematic nursery pollination systems such as figs and fig wasps. Here, we analyse diversification in connection with spatial structuring in the obligate mutualistic association between Ficus septica and its pollinating wasps throughout the and . Results: wasps pollinating F. septica are structured into a set of three vicariant black coloured species, and a fourth yellow coloured species whose distribution overlaps with those of the black species. However, two black pollinator species were found to co-occur on Lanyu island. Microsatellite data on F. septica indicates the presence of three gene pools that broadly mirrors the distribution of the three black clades. Moreover, receptive fig odours, the specific message used by pollinating wasps to locate their host tree, varied among locations. Conclusions: F. septica and its black pollinator clades exhibited similar geographic structuring. This could be due originally to geographic barriers leading to isolation, local adaptation, and finally co-structuring. Nevertheless, the co-occurrence of two black pollinator species on Lanyu island suggests that the parapatric distribution of the black clades is now maintained by the inability of migrating individuals of black pollinators to establish populations outside their range. On the other hand, the distribution of the yellow clade strongly suggests an initial case of character displacement followed by subsequent range extension: in our study system, phenotypic or microevolutionary plasticity has allowed the yellow clade to colonise hosts presenting distinct odours. Hence, while variation in receptive fig odours allows specificity in the interaction, this variation does not necessarily lead to coevolutionary -insect diversification. Globally, our results evidence evolutionary plasticity in the fig-fig wasp mutualism. This is the first documentation of the presence of two distinct processes in pollinating fig wasp diversification on a host species: the formation of vicariant species and the co-occurrence of other species over large parts of their ranges probably made possible by character displacement. Keywords: Biogeography, Ceratosolen, Mutualism, Philippines, Speciation

* Correspondence: [email protected] 1Institute of Biology, University of the Philippines, Diliman, Quezon City, Philippines 2CEFE UMR 5175, CNRS—Université de Montpellier—Université Paul-Valéry Montpellier—EPHE, Montpellier, France Full list of author information is available at the end of the article

© The Author(s). 2017 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Rodriguez et al. BMC Evolutionary Biology (2017) 17:207 Page 2 of 12

Background re-assembly, it has been generally assumed that establish- Interspecific interactions have long been assumed to play ment of obligate mutualisms in islands is highly unlikely an important role in diversification. Competitive coevolu- [15]. Nonetheless, existence of specialized mutualisms in tionary diversification among related species seems to be island complexes defies this assumption. Indeed, the common while mutualistic coevolutionary diversification Glochidion tree-Epicephala moth mutualism is present could be more exceptional [1]. Emblematic examples of throughout the tropical Pacific Islands [15]. Another allopatric speciation followed by competitive character example is the colonization of New Caledonia by section displacement in secondary contact zones include Anoles Oreosycea fig trees and their Dolichoris pollinators [16]. lizards in Caribbean islands [2], Darwin’s finches [3] and Although few, these studies show that dispersal and re- sticklebacks [4]. Nevertheless, despite numerous putative assembly of obligate mutualisms in islands is possible and cases, there are only few fully demonstrated examples of results in evolutionarily stable interactions. Such situations competitive character displacement [5]. allow analysing diversification patterns, some driven by and their pollinators were proposed as models obstacles to gene flow and others by host shifts [15, 17]. that should be prone to mutualistic coevolutionary diver- The intricate island systems of South East Asia and sification, because pollinators disperse the gametes of the Australasia provide a diversity of situations in which Ficus plants whose flowers they visit [6]. Phylogenetic studies species have diversified, and also in which some Ficus spe- actually suggest that reciprocally-associated plant and pol- cies have spread out over numerous islands [18]. Neverthe- linator clades often undergo asynchronous diversification less, little is known about within-Ficus species diversification consistent with a model of plants diversifying in response among islands and about associated pollinating wasp diversi- to sensory biases of their pollinators (e.g., [7]). Further, fication. In insular systems, widespread Ficus species may be microevolutionary studies usually fail to test whether pol- pollinated by different, more or less cryptic species of wasps linators are evolving in response to the plants they visit. In in different parts of their range [17]. This is the case for nursery pollination systems, species-level specificity is Ficus septica, an insular fig species ranging from the Solo- often high, with most plant species usually locally polli- mon Islands to the Ryukyu Islands and throughout insular nated by only one or a few host-specific insect species. South East Asia [17]. Previous work on Ficus septica sug- This suggests that these mutualisms may coevolve. How- gested that coevolutionary diversification through competi- ever, little is known about how speciation and diversifica- tion could be at work among its pollinators, the Ceratosolen tion proceeds in emblematic nursery pollination systems bisulcatus species complex [19]. Further, published data sug- such as figs and fig wasps (but see [8]) or in Phyllantha- gests the presence of among-island variation in receptive fig ceae and their pollinating moths [9]. Because of the intri- odour in Ficus septica, i.e. variation in the main message cate nature of these interactions, they have often been used by the wasps to locate their hosts (compare [19, 20]). proposed as good candidates for diversification through Our goal in this study was to describe the general pattern co-speciation. However, comparing Ficus and their pollin- of diversification within the Ficus septica-pollinating wasps ating wasp phylogenies has shown that only 2/3 of the system and to demonstrate that it is consistent with the nodes are coevolved, and this is the most extreme ex- hypotheses (1) that geographic differentiation of receptive ample documented today of plant-insect codiversification fig odours within Ficus septica,thoughpresent,isnota [10]. At least in some of these systems, pollinator host major biological barrier constraining the distribution of shifts could be a mechanism that promotes diversification pollinator clades, (2) that coevolutionary diversification [1, 11]. In the Yucca-Tegeticula interaction, results suggest through competition followed by expansion through lineage that geographic isolation plays a predominant role in sorting could be involved in an expansion of the Ceratoso- speciation for both Yucca and Tegeticula, with coevolu- len jucundus clade resulting in local coexistence of several tion between them acting primarily to facilitate codiver- pollinator lineages and (3) that the distribution of variation gence after one partner begins to diverge for other reasons in plant and insects does not correspond to a generalized [12, 13]. Similarly in the fig-fig wasp interaction, geo- pattern of diversification through plant-insect reciprocal graphic isolation may also play an important role resulting adaptation as geographic obstacles play a major role. in co-structuring, but does not necessarily result in con- gruent evolutionary histories of plant and insect [14]. Methods While island systems are ideal situations to observe Study system geographic isolation effects, they are also well represented Ficus septica Burm.f. is a small free-standing tree distrib- in demonstrations of coevolutionary diversification as they uted from the Solomon Islands and Northern , provide repetitions of similar events thus providing throughout insular South East Asia, to Taiwan and the statistical power [5]. Moreover, they have limited species Ryukyu Islands. It belongs to subgenus Sycomorus,section diversity allowing related species to diversify into an assort- Sycocarpus and it is functionally dioecious (functionally ment of ecological niches [5]. Because of the difficulty of male trees bear figs that produce pollinating wasp offspring Rodriguez et al. BMC Evolutionary Biology (2017) 17:207 Page 3 of 12

and pollen but no seeds, while female trees bear figs that odours, a highly selected trait of the plants, correlated or produce seeds but neither wasp offspring nor pollen) [18]. not with genetic differentiation in neutral markers of the Two polymorphic species (formally considered as subspe- plants. A complementary aim was to detect whether a given cies) pollinate F. septica, namely Ceratosolen bisulcatus wasp clade could be associated with plant populations exhi- bisulcatus (Mayr) and C. b. jucundus Grandi [21]. Recent biting divergent receptive fig odours, i.e. to detect whether molecular analysis has demonstrated that these species are geographicvariationinreceptivefigodoursconstrainedthe species complexes of closely related species [17], two of distribution of the different wasp clades. which were shown to co-occur in South Taiwan and in Luzon, Philippines, even at the fig-level [19, 22]. Structuring into clades, distribution and phylogeny of Ficus septica pollinators General overview of sampling strategy We sampled pollinator wasps from a large series of sites To get a broad image of the biogeographic distribution of (Table 1, Fig. 1) from March 2011 to May 2014. Nearly F. septica pollinating wasp clades, we collected, sorted to ripe male figs were bagged and wasps were collected from morpho-species and sequenced pollinating wasps from all the bags after their emergence from the figs. The samples over the Philippines, from Taiwan and from Okinawa, and were stored in vials with 70% ethanol. For each location, we examined published molecular data from the same re- wasps were sorted to morpho-species and two to five wasp gion (Table 1). Genetic structuring within F. septica was in- individuals of each morpho-species were sequenced for vestigated on a more limited scale, with four sampling cytochrome c oxidase subunit I (COI), cytochrome b(cyt points in the Philippines (one location in Mindanao, one in b) and elongation factor 1α (EF1α) (following [23, 24]). Negros, two in Luzon), one in Lanyu Island (60 km east of Extractions and amplifications were carried out at CBGP- South Taiwan), and two in Taiwan (Table 1). The aim was INRA, Montferrier-sur-Lez, France. to 1) confirm plant genetic homogeneity within islands and to detect potential genetic differentiation among islands Ficus septica microsatellite analysis and 2) to compare spatial genetic structuring within F. To investigate genetic differentiation of F. septica among septica with the distribution of the different pollinator sites, we sampled leaves from seven study sites between clades. Composition of the odours released by receptive figs April 2010 and September 2011. One leaf was collected (i.e. figs ready to be visited by pollinators), was investigated from 17 to 37 trees per site in the Philippines and Taiwan in four locations, one in Mindanao, one in Negros, one in (Table 1). Two pairs of collection sites were located on the LuzonandoneinTaiwan,soastorepresentthezonesin- same islands: on Luzon, they were about 100 km apart vestigated for plant genetic structuring, and for which we and on Taiwan, they were about 350 km apart. The leaves also had information on pollinating wasps (Table 1). The were oven-dried for two to three days and preserved dry. aim was to detect whether differentiation in receptive fig DNA was extracted from the leaf samples, amplified and

Table 1 Collection sites for pollinator, leaf and fig odour samples Country Locality F. septica pollinators F. septica leaves F. septica receptive fig odour Japan Okinawa [55] Japan Okinawa This study Taiwan North This study, [17, 23] This study This study Taiwan Throughout the country [56] Taiwan South This study, [17, 22] This study Taiwan Lanyu Island [17, 22, 56] This study Philippines Northern Luzon This study Philippines Central Luzon This study This study This study Philippines Southern Luzon This study, [57] This study Philippines Panay Island This study Philippines Negros Island This study This study This study Philippines Palawan Island This study Philippines Camiguin Island This study Philippines Mindanao Island This study This study This study Table 1 Legend: Sites are arranged in a north-to-south orientation. Also indicated are the references of the GenBank sequences (Additional file 8) that were blasted on our sequences for clade assignation. Pollinator genes analysed - [17, 50–52]: COI;[48]: COI, 28S;[49]: 18S, 28S, COI, cytb, Wg; this study: COI, cyt b, EF1α. The COI sequences allow the assignment of every sequenced individual to clade and subclade recognised in this study Rodriguez et al. BMC Evolutionary Biology (2017) 17:207 Page 4 of 12

Fig. 1 Geographic distribution of the Ficus septica gene pools and the different clades (and subclades) of the pollinators of F. septica

genotyped (following [25]). Fourteen nuclear microsatel- Ficus septica receptive fig odour collection and analysis lite markers specifically designed for the genus Ficus were To collect odours from receptive figs, 5–20 F. septica trees used [25]. They were selected because of their high were sampled per study site from March to September amplification rates and polymorphism (Additional file 1). 2011. Scent extraction was carried out from 10.00 to 14.00 Genotyping was carried out using an ABI 3730 Sequencer when scent emission is maximum [19]. As much as pos- (Applied Biosystems). sible, male and female trees were equally represented in Rodriguez et al. BMC Evolutionary Biology (2017) 17:207 Page 5 of 12

the sampling. Moreover, in the Philippines, fig scent col- We used the program STRUCTURE PLOT to visualize the lection was performed twice per site, at different seasons. Bayesian clusters [32]. Odours were collected using the headspace technique. Significant differences in total quantity of volatiles were Thirty to seventy receptive figs from each of the selected tested using One-way ANOVA. For the subsequent analyses, trees were placed inside a polyethylene terephthalate bag. relative abundance data was used. Fig scent composition The bag was closed for 30 min to allow the scent to accu- was analyzed by constructing several models of MANOVA mulate. After the accumulation period, air was pumped out using geography, sex and season (or combinations of these) of the bag through two tubes, each with a chromatoprobe as explanatory variables. Significance values of each response − filter, for 5 min at a rate of 158 mL min 1. The volatile com- variable were evaluated using the false discovery rate [33]. pounds were adsorbed into these filters, which were previ- Spread of the variation in the data was visualized in a ously injected with known amounts of standard compounds principal components analysis (PCA) performed using the (nonane and dodecane) to allow quantification of the vola- package FACTOMINER v1.31.5. A dissimilarity matrix using tiles. Empty bags were used as controls. The filters were the Bray-Curtis index was also generated. Inter-population placed in vials and stored at −18 °C until further analysis. comparison of the blends was done by subjecting the Bray- To identify and quantify the compounds present in the Curtis indices to permutational analysis of variance (perma- scent samples, filters were injected in a gas chromato- nova) tests (permutations = 100,000) in the package VEGAN graph (CP-3800, Varian Inc., Palo Alto, CA) coupled with v2.3–3. All analyses were performed in R v3.2.1. a mass spectrometer (Saturn 2000, Varian; for identifica- tion of compounds) and equipped with a flame ion de- Results tector (for quantification of compounds). Samples were Ficus septica microsatellite analysis injected using a 1079 programmed temperature injector Genetic parameters averaged over 14 Ficus septica micro- with a chromatoprobe thermodesorption kit (Varian). satellite loci used in this study are shown in Additional Temperature program and other conditions followed [19]. files 1 and 2. Expected heterozygosity (H ) and number of Component identification was based on the gas chromato- e alleles per locus (Na) exhibited minimal variation in the graph (GC) retention indices of the compounds compared seven sites (0.313–0.405) while private alleles (PA) varied to those reported in literature [26] and through computer more between islands. Exact tests of genotypic differenti- matching of mass spectra (MS) with NIST’98 MS library. ation showed that all the population pairs, except the two These analyses were performed at the Plate-Forme d’Ana- from Luzon Island, were differentiated. lyses Chimique en Ecologie of the LabEx CeMEB. FST values showed that the least differentiated populations were those from Luzon and those from Taiwan and Lanyu Data analyses (Additional file 3). An intermediate homogeneity between For the phylogenetic analysis of pollinator wasp sequences, the Luzon populations and the Negros population was gene sequences were cleaned and edited in Geneious v5.5.9 evidenced by intermediate F values demonstrating some and checked for presence of pseudogenes in Mega v6.06. ST genetic differentiation. The Bayesian clustering gave an opti- We ran maximum likelihood (ML) analyses on the mito- mal number of clusters, K, equal to 3 (Fig. 2). From South chondrial DNA dataset (COI - 1527 bp + cyt b -734bp) to North, the clusters gathered 1) the lone Mindanao popu- and on the nuclear DNA dataset (EF1a - 517 bp) using a lation, 2) the Negros and Luzon populations, 3) the Lanyu GTR + Γ4 model with 1000 bootstrap replicates. We had Island population and the two Taiwan populations (Fig. 2). 54 COI,65cyt b and 58 EF1α in-group sequences. We also used sequences of pollinator wasps of F. botryocarpa, F. malayana and F. uncinata as outgroups. In total, we had Ficus septica receptive odours 13 COI,10cyt b and 8 EF1α outgroup sequences. A total of 39 compounds were found in the scent of F. sep- To quantify genetic variation between the F. septica popu- tica, with a little less than half classified as sesquiterpenes lations, heterozygosity, mean number of alleles, and FST (Additional file 4). However, the two dominant compounds, values were computed using GENALEX v6.5 [27] and SPAGEDI (E)-β-ocimene and (Z)-3-hexen-1-ol, were not sesquiter- V1.4 [28]. Exact tests of inter-population genetic differenti- penes but an acyclic monoterpene and a simple aliphatic, ation were carried out using Fisher’smethodinGENEPOP respectively. The most abundant fig odour component was v4.2 [29]. Finally, the microsatellite data was subjected to α-pinene in Mindanao samples, (Z)-3-hexen-1-ol in Negros Bayesian clustering analysis using STRUCTURE v2.3.4 [30] in samples, (E)-β-ocimene in Central Luzon samples, and order to determine the different genetic groups. The param- linalool in North Taiwan samples. Total amount of volatiles eter set was a burn-in period of 100,000 steps and a run of emitted did not differ significantly between locations, 1,000,000 steps with no admixture, repeated five times. The seasons and sexes (One-way ANOVA, F10,45 =1.853, number of individual clusters was determined using the P = 0.078). Of the 39 compounds, 18 had a mean amount Evanno method implemented in STRUCTURE HARVESTER [31]. of at least 1% of the blend (Additional file 4). To allow Rodriguez et al. BMC Evolutionary Biology (2017) 17:207 Page 6 of 12

Fig. 2 Plant genetics. Bar plots of the Bayesian clustering runs at the best K (3). Each column displays the probability of each individual to belong to each of the clusters. The three groups corresponded to i) Taiwan including Lanyu Island, ii) Luzon plus Negros Island, iii) Mindanao Island significance testing, only the abundances of these 18 com- geography/location is the main factor affecting fig scent pounds were used for subsequent analyses. variation, with the sex of the tree contributing very little, if Using relative abundance values of the 18 compounds, at all, to the variation seen. Lastly, in the Philippines, season we found that geography/location, season and their inter- also accounted to some extent for scent variation in Negros action factor were significant in explaining the variation in and Central Luzon but not in Mindanao. odour blend of receptive figs from Philippines (permanova, geography: F = 6.93, P < 0.0001; season: F = 2.55, Structuring into clades and phylogeny of Ficus septica P = 0.016; geography*season: F = 1.896, P = 0.026). Sex did pollinators not have a significant effect on scent variation (permanova, Number of individuals sequenced, location of collection and sex: F = 0.49, P = 0.868). When the Philippine dataset was GenBank accession numbers are given in Additional file 8. partitioned into the three different sites, season had a sig- ThephylogenetictreesderivedfromnuDNA(Fig.4A)and nificant effect in the scent variation in both Negros and mtDNA (Fig. 4B) show the presence of four major clades as- Central Luzon, but not in Mindanao (permanova: Negros sociated with F. septica. In our samples, three major pollin- – F = 2.60, P = 0.012; Central Luzon – F = 2.89, P = 0.021; ator clades (1–3) corresponded to black coloured morphs Mindanao – F =1.20,P = 0.30; Additional file 5). In an and the last major clade (clade 4) was composed of more or analysis of all samples, including Taiwan, geography/loca- less yellow-coloured morphs, either dark backed-yellow tion still showed a significant effect and sex still presented sided (subclades 4A and 4B) or totally yellow (subclades 4C no effect on the blend (MANOVA, geography: approx. and 4D). Clades 1 to 3 followed a south to north distribu- F=5.45, P < 0.0001; sex: approx. F =0.44,P = 0.967). tion: clade 1 was present only in Mindanao, clade 2 ranged Through pairwise comparisons of samples from different from Negros Island to Luzon (plus, according to COI se- locations, it was found that the odour blend of the four F. quences, one specimen collected on Lanyu Island and ana- septica populations significantly differed from each other lyzed in [22]), and clade 3 occurred from Lanyu Island to (Additional file 6). This result was supported by the find- Okinawa Island (Fig. 1). On the other hand, clade 4 showed ings that the main compounds in the blends of the four less geographic structuring and was present throughout the sites differed (Additional file 7). PCA also reflected this dif- Philippines up to South Taiwan (Fig. 1). Subclade 4A ferentiation (Fig. 3). The first two PCs explain about 24% occurred in Mindanao and neighboring Camiguin Island, of the variation in the data. The first axis separated the subclade 4B stretched from Palawan to the southern tip of Mindanao samples from the other samples. The second Luzon. Subclade 4C ranged from Central to North Luzon axis separated the remaining three groups. In this repre- and subclade 4D was present in Lanyu Island and South sentation, the barycenters of fig odours from Central Taiwan. Clade 1 was found in sympatry with subclade 4A. Luzon and Negros were located next to each other. Clade 2 coexisted with subclade 4C (Luzon) and with In summary, data on receptive fig odour show that Min- subclade 4D and clade 3 (Lanyu Island). Lastly, subclade 4B danao, Negros, Central Luzon, and North Taiwan each have was not found in sympatry with any other clade. All previ- their own distinct odour profiles. These also show that ously published sequences from our sampling region fitted Rodriguez et al. BMC Evolutionary Biology (2017) 17:207 Page 7 of 12

into the picture presented here. Samples from Indonesia and New Guinea [17] belonged to different clades from those recovered in this study.

Discussion Genetic and odour structuring in Ficus septica F. septica was spatially structured into three separate gene pools (from South to North): 1) the Mindanao population, 2) the Negros and Luzon populations, 3) the Lanyu Island population and the two Taiwan populations (Fig. 2). How- ever, this genetic structuring was not reflected by the recep- tive fig odours because Mindanao, Negros, Central Luzon, and North Taiwan each had their own distinct odour profiles (Fig. 3). Hence receptive fig odours seem to evolve differently from neutral markers, suggesting selection. Be- yond the case of F. septica, geographic variation in receptive fig odours has also been demonstrated in two fig species, F. Fig. 3 Plant odours. Principal Components Analysis of relative hispida and F. racemosa.AtleastinF. racemosa,receptive abundance of each compound in receptive fig odour showing fig odours varied between plants belonging to different gene groupings according to location. (North Taiwan – Yellow, Central pools and pollinated by different wasp species, while recep- Luzon – Red, Negros Island – Blue, Mindanao Island – Green). Odour profiles vary significantly between each of the four study sites tive fig odours did not differ between plants growing (MANOVA, geography: approx. F=5.45, P < 0.0001) 800 km apart but belonging to the same gene pool and pollinated by the same wasp species [14, 34]. This similarity

Fig. 4 Insect phylogeny. Maximum likelihood (ML) trees from (a) nuclear DNA (EF1α gene sequences) and (b) mitochondrial DNA (COI and cyt b gene sequences) showing the four pollinator clades (clades 1 to 4) locally-associated with Ficus septica. Clade 4 in (b) is further subdivided into subclades 4A to 4D Rodriguez et al. BMC Evolutionary Biology (2017) 17:207 Page 8 of 12

in fig odours suggests stabilising selection at work. Further, undescribed species that all appear different from C. in the exceptional situation where two Ficus species freely bisulcatus (stat. nov.). share pollinators, their receptive fig odours are not distin- The three black coloured clades were allopatric to parapa- guishable, suggesting selection by the wasps for evolution- tric, presenting a geographic South-North zonation (Fig. 1). ary convergence of receptive fig odours [35, 36]. While we The black clades presented a zonation similar to that of the donotprovideadirectdemonstrationthatthedifferencein F. septica gene pools (Fig. 2). While sampling was not receptive fig odour reported here is meaningful for pollinat- strictly comparable preventing formal statistical comparison, ing wasp attraction, more rapid evolution than neutral the similarity in spatial structure of clade distributions was markers among islands is strikingly different from the striking. We hypothesize that this co-structuring is due to stability and convergence of receptive fig odours docu- thepresenceofgeographicbarriers between the populations mented in other Ficus species. Therefore we hypothesise leading to isolation and subsequent diversification. F. septica that the differentiation of receptive fig odours in our study and the black pollinator clades might have colonized the system is meaningful in terms of pollinator attraction. islands along the same South-to-North direction. Upon The importance of receptive fig odours in pollinating establishment in the different islands, populations must wasp attraction has been largely demonstrated [37–39]. In have been geographically isolated, subsequently facili- contrast, variation within fig species [34] and among closely tating local adaptation and diversification, such that the related species has been much less investigated [35, 36]. co-structuring between the fig and the wasps is evident We document here seasonal variation in receptive fig now. Co-structuring, in this case, is thus driven more odours in Luzon and Negros, but not in Mindanao, a differ- by geographic (and accompanying climatic) differences ence that may be explained by limited seasonality in Min- rather than by strict plant-insect reciprocal adaptation. danao comparatively to the two other islands [40]. This fits Indeed, according to the Köppen-Geiger climate classi- the very limited available data showing seasonal variation in fication, the climates of Luzon and Negros are similar receptive fig odours, measured within or between sexes de- while the climates of Mindanao and Taiwan are more pending on the species’ flowering phenology [20, 41]. different [40]. This climatic differentiation is in broad Nevertheless, in all documented cases, differences between concordance with the distributions of pollinator clades seasons remain limited, more limited than differences 1to3andF. septica gene pools. among species (except for Ficus species sharing pollinators How do these patterns relate to receptive odour structur- [35, 36]) or than differences among islands for F. septica. ing? If an olfaction barrier maintains the specificity of the fig- wasp interactions, then we would expect each pollinator Spatial structuring in Ficus septica and its pollinator clade to recognize a single receptive odour profile. Neverthe- clades less, geographic variation in receptive fig odours in F. septica In F. septica, we show that gene pools do not neatly co- has not precluded the presence of the following three situa- structure with receptive fig odours, as one gene pool tions: (1) coexistence of clades 2 and 3 on Lanyu Island, (2) (Luzon-Negros) corresponded to two distinct scent profiles presence of clade 2 in Luzon and Negros despite differing fig (Luzon and Negros). This difference provides us with a new scent, and if our scenario is correct, (3) geographic expansion perspective to analyse pollinator structuring. Will pollinator of clade 4 throughout the Philippines and up to Taiwan, species/populations reflect the geographic structuring of F. bridging the different receptive fig odours produced by F. septica gene pools? Or will pollinators co-structure with septica. These suggest either that wasps can rapidly evolve receptive odour profiles, which are the main wasp attract- detection of new odours or that the wasps can still ing agents? recognize receptive figs because the odour differences are The pollinators of Ficus septica formed a monophyletic limited. Hence, while receptive fig odour is a cue wasps entity, structured into four major genetic clades that can be use to locate their host figs [39], its differentiation among recognized morphologically, and four subclades within fig populations or among related Ficus species may allow clade 4 that could not be distinguished morphologically, to phenotypic or micro-evolutionary plasticity in wasp odour the exception of colouration differences. According to avail- recognition rather than being a constraint that would lead able data on genetic distances within and among fig pollin- to strict insect-plant coevolutionary diversification. We ating wasp species, the four major clades should be may conjecture, for instance, that response to receptive fig considered as distinct species while the subclades should be odours, and maybe even perception of odours in wasps of conservatively interpreted as within species variation [42]. clade 2 may vary between wasps originating from Negros, ThetypeofCeratosolen jucundus was sampled on Mount from Luzon and from Lanyu Island. Future experimental Makiling [43], where we collected subclade 4C. Within this studies are needed to establish possible differential attrac- framework, Ceratosolen bisulcatus jucundus corresponds to tion to the various receptive odour compounds. clade 4 and is upgraded to species rank Ceratosolen jucun- We therefore demonstrate here that structuring in F. dus (stat. nov.), while clades 1 to 3 correspond to three septica and their pollinator clades is driven mainly by Rodriguez et al. BMC Evolutionary Biology (2017) 17:207 Page 9 of 12

geographic (and climatic) factors rather than a species- would qualify as ecological character displacement sensu specific olfaction barrier. stricto, while subsequent range extension would be made possible by differences evolved in allopatry, in a process of Pollinator clade distribution and diversification species sorting [5]. It should be noted that clades 2 and 3 are both present Interestingly, our results show the widespread co- in Lanyu Island, which could be considered as their con- occurrence of yellow C. jucundus wasps with the black tact zone. Although these two clades are distributed over pollinator clades. This has been previously reported, but large stretches of sea presenting only few small islands, on a much smaller scale [19, 22, 45]. Co-existence of re- they still retain within-clade genetic homogeneity. This productively isolated yellow and black pollinator clades suggests sufficient effective numbers of migrants and were also documented in F. rubiginosa collected from hence sufficient dispersal. How then is the co- Townsville, Australia [42, 46]. Aside from these studies structuring maintained? What prevents expansion of and a few more [47], co-occurrence of sympatric, actively clade 2 into South Taiwan (a distance of about 65 km pollinating species over large areas has been rarely docu- from Lanyu Island) when it is able to hop along a series mented and their population genetics barely characterized. of islands up to 100 km apart from Luzon to Lanyu? Our study fills this gap and we have proposed mechanisms Our working hypothesis is that migrant pollinators are on how this co-existence was achieved. unable to establish viable populations outside their real- ized range. This might be because local pollinators are Biogeography and adaptation in relation to other more adapted to local conditions than the migrants, fig-wasp systems leading to competitive exclusion of migrating species. How do our results compare with what is known about This pattern was also exhibited by the non-pollinating biogeography and adaptation in the fig-pollinating wasp fig wasps associated with Ficus rubiginosa in Australia, system? The situation in F. racemosa resembles the situ- where parapatric species pairs had smaller ranges than ation of F. septica and its black pollinator clades, but on a expected probably due to competition [44]. Therefore, broader scale. Indeed, the pollinators are structured into a the parapatric distribution of the black clades might be set of wide-ranging vicariant species and the plant is struc- maintained by the inability of migrating species to estab- tured along the same geographic lines into different gene lish populations outside their range. pools, the limit of which seem to correspond to geographic Within this context, the distribution and incipient diversi- barriers limiting gene flow [14]. Similarly, results on F. hirta fication of clade 4 is of particular interest. Clade 4 is present and its pollinator in South East China show overall strong throughout the Philippines and extends 60 km northwards gene flow over continental expanses in both plant and in- into South Taiwan, bridging almost the whole set of cli- sect, with genetically divergent populations for both plant mates. However, the distribution of subclades is structured and insect on Hainan island [48]. The authors suggested geographically with a South-to-North zonation. Also, sub- that the geographic barrier in itself was not sufficient to ex- clades 4C and 4D are characterized by completely yellow plain the data, and that isolation by adaptation, facilitated colour in all the specimens we collected and their distribu- by a decrease in dispersal across a geographic barrier, had tions totally overlap with the distributions of black coloured to be invoked [48]. In the data presented here, the presence clades. Yellow colour is exceptional in diurnal fig pollinator of clade 2 in Lanyu Island supports this vision, as it shows species [19]. Individuals of subclade 4C have been demon- that fig pollinating wasp species can achieve continuity strated to exhibit very short survival times outside figs com- across almost 400 km of sea, hopping from island to island paratively to co-occurring individuals of the black coloured with some over 100 km apart: the 30 km separating Hainan clade 2 [19]. The authors interpreted the yellow colour of island from the continent can hardly be considered by the clade as part of an adaptive syndrome based on short themselves a major obstacle for fig wasp dispersal. More- survival but high competitiveness, a “live fast, die young” over, data on F. pumila, a creeping fig, and its pollinators in strategy. Hence, clade 4 seems to have evolved a modified Eastern China showed some genetic isolation by distance in life strategy, which has enabled colonization of areas where the plant [49], and separation of the pollinators into two F. septica is also pollinated by a black coloured clade and wide ranging vicariant species. South of Shanghai, in the the subsequent stable coexistence of the two wasp species. Zhoushan archipelago, within a context of some geographic This suggests recent range expansion into regions inhabited obstacles to dispersal, a localized third pollinator species is by clades 1 to 3 and ongoing differentiation into distinct present in combination with one of the wide-ranging gene pools. Such a scenario would involve initial ecological species [50]. The two co-occurring species differ in preco- character displacement in a region of co-occurrence city of emergence from figs after over-wintering, suggesting followed by subsequent colonization of the regions inhab- differences in temperature thresholds. This ecological ited by the other black clades. In this sense, the initial clade differentiation may facilitate the coexistence of competing 4 colonization of habitats already occupied by a black clade species [50], while the parapatric distribution of the two Rodriguez et al. BMC Evolutionary Biology (2017) 17:207 Page 10 of 12

wide-ranging species is probably best explained by competi- adaptation, and finally plant-insect co-structuring. We also tive exclusion in a context of strong gene flow limiting local show the co-existence of pollinating species in an area, adaptation in widespread species [51]. Finally, F. rubiginosa which probably had been brought about by initial character provides the most complex pattern of distribution of displacement and subsequent range expansion. This is the diverse pollinator species documented to date, a feature first documentation of the presence of two distinct that may result from the range of the host tree encompass- processes in pollinating fig wasp diversification on a host ing a combination of two climatic gradients, a progressive species: the formation of vicariant species and the co- North-South gradient and a steep East-West one [42, 46, occurrence of other species over large parts of their ranges 52]. Hence, the biogeographic patterns observed in fig- probably made possible by character displacement. Lastly, pollinating wasp associations tend to support a global vision we provide evidence that receptive fig odours do not con- in which the pollinators of wide ranging fig species, but not strain the specificity of the fig-pollinator wasp interactions. necessarily the fig species, are often structured into broad, geographically vicariant, genetically homogeneous entities. Additional files New lineages of pollinating wasp species may colonize such systems following a local event of ecological character Additional file 1: Characteristics of 14 microsatellite loci used in the displacement. This can be followed by subsequent study across all Ficus septica individuals. Ho - observed heterozygosity, He - expected heterozygosity, Na - number of alleles per locus, NA - Not colonization of parts of the range of the fig occupied by Applicable. (DOCX 61 kb) other pollinators through a process of species sorting Additional file 2: Genetic parameters from 14 microsatellite loci of Ficus – resulting in the regional coexistence of two pollinating septica. n number of leaves sampled, He - expected heterozygosity, Na - mean wasps on the same host. We were able to demonstrate number of alleles per locus, PA - mean number of private alleles. (DOCX 51 kb) these general emergent patterns for pollinating fig wasp Additional file 3: Genetic differentiation in Ficus septica populations as shown from FST values computed from 14 microsatellite loci. All FST species diversification in a relatively simple fig-wasp system. values (besides between the two populations from Luzon) are significant This leads us to suggest that within Ficus species, in an in- at P < 0.05 using the individual permutation tests. TW – Taiwan; PH – sular setting, fig-fig wasp diversification and co-structuring Philippines. (DOCX 62 kb) may be frequent. These patterns do not need to result from Additional file 4: Volatile organic compounds emitted by Ficus septica from the Philippines and North Taiwan. Compounds with more than 1% coevolutionary diversification[1].Indeed,inourfig-fig mean relative abundance are shown in boldface. (DOCX 79 kb) wasp system, geographic barriers seem to have been the Additional file 5: Plot of the mean relative abundance of the 18 most main driving factor leading to plant-insect co-structuring, a abundant compounds found in Ficus septica odours between two seasons (dry feature that does not necessarily entail parallel cladogenesis and rainy) in two Philippine sites: Central Luzon and Negros Island. (PDF 9 kb) [14]. Interestingly, the co-occurrence of clades 2 and 3 in Additional file 6: Permutational analysis of variance (permanova) comparisons of the fig odour profiles (relative abundance) of Ficus septica Lanyu Island suggests that barriers that might arise from populations from North Taiwan and three Philippine sites (Central Luzon, coevolutionary diversification can be overcome, limiting its Negros Island, Mindanao Island) (*-significant difference at P < 0.05). role in the process of joint plant-insect diversification. Significant P-values for all comparisons show that each of the 4 sites has its own distinct odour profile. (DOCX 56 kb) The Philippines constitute a model island archipelago to Additional file 7: Plot of the mean relative abundance of the 18 most investigate evolutionary processes of diversification [53]. abundant compounds found in Ficus septica odours from North Taiwan and Much focus has been on limited dispersal species, for three sites in the Philippines (Central Luzon, Negros Island, Mindanao Island): which routes of colonization can be reconstructed using the most abundant compound is different in each of the sites. (PDF 9 kb) phylogenies. The results presented here on Ficus septica Additional file 8: Sampling table for Ceratosolen bisulcatus wasps indicating collection locality and GenBank accession number for each allow a different perspective. What biogeographic structure sequence used for the phylogenetic analysis. (DOCX 108 kb) do we get in strong dispersal model species and what does it tell us about evolutionary processes at work in lower Abbreviations dispersal species? Our study may shed light on results such ANOVA: analysis of variance; COI: cytochrome c oxidase subunit I; cyt α α as those on shrews [54] as it supports a role for geographic b: cytochrome b; EF1 : elongation factor 1 ; GC: gas chromatograph; GTR: general time reversible; He: expected heterozygosity; MANOVA: multivariate analysis of barriers, but probably also for ecological character variance; ML: maximum likelihood; MS: mass spectra; mtDNA: mitochondrial DNA; displacement followed by species sorting in diversification Na: number of alleles per locus; nuDNA: nuclear DNA; PA: private alleles; processes. Our results also emphasize the importance of PC: principal component; PCA: principal components analysis; Permanova: permutational analysis of variance climatic variation among islands as a force structuring diversity and diversification. Acknowledgements We are grateful to Jean-Marie Bessiere, Marjorie Garcia, Bruno Buatois, Florian Massip, Rian Bernal, Nonoy Dulman, Luciano Palmieri, Ian Fontanilla, Dino Conclusions Ramos, Yoyi Rodriguez and Reynand Canoy for helping us in the We have shown that F. septica and its black pollinator identification of compounds, laboratory work and field collection. clades exhibited similar geographic structuring. We Funding hypothesize that this could be due originally to strong This work was supported by the University of the Philippines Office of the geographic (and climatic) barriers leading to isolation, local Vice-Chancellor for Research and Development, Commission on Higher Rodriguez et al. BMC Evolutionary Biology (2017) 17:207 Page 11 of 12

Education (Philippines), ANR-09-BLAN-0392-CSD 7, NSC 99–2923-B-002-001- 13. Yoder JB, Smith CI, Rowley DJ, Flatz R, Godsoe W, Drummond C, et al. MY3, Bibliothèque du Vivant funded by the CNRS, Muséum National d’Histoire Effects of gene flow on phenotype matching between two varieties of Naturelle and the Institut National de la Recherche Agronomique. Joshua tree (Yucca brevifolia; Agavaceae) and their pollinators. J Evolution Biol. 2013;26:1220–33. Availability of data and materials 14. Bain A, Borges RM, Chevallier MH, Vignes H, Kobmoo N, Peng YQ, et al. Microsatellite genotyping data, odour profiles and gene alignments are Geographic structuring into vicariant species-pairs in a wide-ranging, high- available from the corresponding author on request. Wasp DNA sequences are dispersal plant-insect mutualism: the case of Ficus racemosa and its deposited in GenBank (accession numbers are listed in the Additional file 8). pollinating wasps. Evol Ecol. 2016;30:663–84. doi:10.1007/s10682-016-9836-5. 15. Hembry DH, Okamoto T, Gillespie RG. Repeated colonization of remote Authors’ contributions islands by specialized mutualists. Biol Lett. 2011;8:258–61. MHM, LSC, JYR and FK conceived the study. LJR, AB, LC, AC, RG, HYT, JYR and FK 16. Cruaud A, Rasplus JY, Jabbour-Zahab R, Genson G, Ungricht S. Testing the performed the experiments and data collection. LJR, AB, LC, AC, JYR and FK emergence of New Caledonia: fig wasp mutualism as a case study and a analysed and interpreted the data. LJR and FK wrote the manuscript. All authors review of evidence. PLoS One. 2012;7:e30941. commented on the manuscript and gave final approval for publication. 17. Moe AM, Weiblen GD. Molecular divergence in allopatric Ceratosolen (Agaonidae) pollinators of geographically widespread Ficus () Ethics approval and consent to participate species. Ann Entomol Soc Am. 2010;103:1025–37. We confirm that the leaf and insect collection presented here were conducted 18. Berg CC, Corner EJH. Moraceae (Ficus). In: HPNooteboom, editor. Flora in accordance with the Wildlife Resources Conservation and Protection Act Malesiana. Netherlands: Nationaal Herbarium Nederland, Universiteit Leiden (Republic Act 9147) set forth by the Philippine government, the Department of branch. 2005. p. 1–730. Environment and Natural Resources and the University of the Philippines. 19. Conchou L, Cabioch L, Rodriguez LJV, Kjellberg F. Daily rhythm of mutualistic pollinator activity and scent emission in Ficus septica: ecological Consent for publication differentiation between co-occurring pollinators and potential Not applicable consequences for chemical communication and facilitation of host speciation. PLoS One. 2014;9:e103581. Competing interests 20. Hossaert-McKey M, Proffit M, Soler C, Chen C, Bessière JM, Schatz B, et al. The authors declare that they have no competing interests. How to be a dioecious fig: chemical mimicry between sexes matters only when both sexes flower synchronously. Sci Rep. 2016;6:21236. Publisher’sNote 21. Wiebes JT. Taxonomy and host plant preference of indo-Australian fig wasps of the genus Ceratosolen (Agaonidae). Tijdschr Entomol. 1963;106:1–112. Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. 22. Lin RC, Yeung CKL, Fong JJ, Tzeng HY, Li SH. The lack of pollinator specificity in a dioecious fig tree: sympatric fig-pollinating wasps of Ficus – Author details septica in southern Taiwan. Biotropica. 2011;43:200 7. 1Institute of Biology, University of the Philippines, Diliman, Quezon City, 23. Cruaud A, Jabbour-Zahab R, Genson G, Cruaud C, Couloux A, Kjellberg F, et Philippines. 2CEFE UMR 5175, CNRS—Université de Montpellier—Université al. Laying the foundations for a new classification of Agaonidae Paul-Valéry Montpellier—EPHE, Montpellier, France. 3INRA, UMR 1062 CBGP, (hymenoptera: Chalcidoidea), a multilocus phylogenetic approach. Cladistics. – Montferrier-sur-Lez, France. 4Institute of Ecology and Evolutionary Biology, 2010;26:359 87. National Taiwan University, , Taiwan. 5Institut d’Ecologie et des Sciences 24. Cruaud A, Jabbour-Zahab R, Genson G, Couloux A, Peng YQ, Yang DR, et al. Out de l’Environnement de Paris—ECOSENS, INRA-UPMC, Versailles, France. of Australia and back again: the world-wide historical biogeography of non- – 6Department of Forestry and Natural Resources, National Chung-Hsing pollinating fig wasps (hymenoptera: Sycophaginae). J Biogeogr. 2011;38:209 25. University, Taichung, Taiwan. 25. Garcia M, Bain A, Tzeng HY, Peng YQ, Chou LS, Kjellberg F. Portable microsatellite primers for Ficus (Moraceae). Am J Bot. 2012;99:e187–92. Received: 17 March 2017 Accepted: 4 August 2017 26. Adams RP. Identification of essential oil components by gas chromatography/mass spectroscopy. 4th ed. USA: Allured Publishing Corporation; 2007. References 27. Peakall R, Smouse PE. GenAlEx 6.5: genetic analysis in excel. Population — 1. Hembry DH, Yoder JB, Goodman KR. Coevolution and the diversification of genetic software for teaching and research an update. Bioinformatics. – life. Am Nat. 2014;184:425–38. 2012;28:2537 9. 2. Stuart YE, Campbell TS, Hohenlohe PA, Reynolds RG, Revell LJ, Losos JB. 28. Hardy OJ, Vekemans X. SPAGeDi: a versatile computer program to analyse Rapid evolution of a native species following invasion by a congener. spatial genetic structure at the individual or population levels. Mol Ecol Science. 2014;346:463–6. Notes. 2002;2:618–20. 3. Grant PR, Grant BR. How and why species multiply: the radiation of Darwin's 29. Rousset F. GENEPOP'007: a complete re-implementation of the GENEPOP finches. New Jersey: Princeton University Press; 2008. software for windows and Linux. Mol Ecol Resour. 2008;8:103–6. 4. Schluter D, McPhail JD. Ecological character displacement and speciation in 30. Pritchard JK, Stephens M, Donnelly P. Inference of population structure sticklebacks. Am Nat. 1992;140:85–108. using multilocus genotype data. Genetics. 2000;155:945–59. 5. Stuart YE, Losos JB. Ecological character displacement: glass half full or half 31. Earl DA. vonHoldt BM. STRUCTURE HARVESTER: a website and program for empty? Trends Ecol Evol. 2013;28:402–8. visualizing STRUCTURE output and implementing the Evanno method. 6. Thompson JN. The coevolutionary process. Chicago: The University of Conserv Genet Res. 2011;4:359–61. Chicago Press; 1994. 32. Ramasamy RK, Ramasamy S, Bindroo BB, Naik VG. STRUCTURE PLOT: a 7. Ramirez SR, Eltz T, Fujiwara MK, Gerlach G, Goldman-Huertas B, Tsutsui ND, program for drawing elegant STRUCTURE bar plots in user friendly interface. et al. Asynchronous diversification in a specialized plant-pollinator SpringerPlus. 2014;3:431. mutualism. Science. 2011;333:1742–6. 33. Benjamini Y, Hochberg Y. Controlling the false discovery rate: a practical and 8. Yokoyama J. Cospeciation of figs and fig-wasps: a case study of endemic powerful approach to multiple testing. J Roy Stat Soc B Met. 1995;57:289–300. species pairs in the Ogasawara Islands. Popul Ecol. 2003;45:249–56. 34. Soler C, Hossaert-McKey M, Buatois B, Bessière JM, Schatz B, Proffit M. 9. Kawakita A. Evolution of obligate pollination mutualism in the tribe Geographic variation of floral scent in a highly specialized pollination Phyllantheae (Phyllanthaceae). Plant Spec Biol. 2010;25:3–19. mutualism. Phytochemistry. 2011;72:74–81. 10. Cruaud A, Rønsted N, Chantarasuwan B, Chou LS, Clement WL, Couloux A, 35. Cornille A, Underhill JG, Cruaud A, Hossaert-McKey M, Johnson SD, Tolley et al. An extreme case of plant-insect codiversification: figs and fig- KA, et al. Floral volatiles, pollinator sharing and diversification in the figwasp pollinating wasps. Syst Biol. 2012;61:1029–47. mutualism: insights from Ficus natalensis, and its two wasp pollinators 11. Cook JM, Segar ST. Speciation in fig wasps. Ecol Entomol. 2010;35:54–66. (South Africa). Proc R Soc B. 2012;279:1731–9. 12. Kiester AR, Lande R, Schemske DW. Models of coevolution and speciation in 36. Wang G, Cannon CH, Chen J. Pollinator sharing and gene flow among closely plants and their pollinators. Am Nat. 1984;124:220–43. related sympatric dioecious fig taxa. Proc R Soc B. 2016;283:20152963. Rodriguez et al. BMC Evolutionary Biology (2017) 17:207 Page 12 of 12

37. Hossaert-McKey M, Gibernau M, Frey JE. Chemosensory attraction of fig wasps to substances produced by receptive figs. Entomol Exp Appl. 1994;70:185–91. 38. Grison-Pigé L, Bessière JM, Hossaert-McKey M. Specific attraction of fig- pollinating wasps: role of volatile compounds released by tropical figs. J Chem Ecol. 2002;28:267–79. 39. Chen C, Song Q, Proffit M, Bessière JM, Li Z, Hossaert-McKey M. Private channel: a single unusual compound assures specific pollinator attraction in Ficus semicordata. Funct Ecol. 2009;23:941–50. 40. Peel MC, Finlayson BL, McMahon TA. Updated world map of the Köppen- Geiger climate classification. Hydrol Earth Syst Sci. 2007;11:1633–44. 41. Soler CC, Proffit M, Bessiere JM, Hossaert-McKey M, Schatz B. Evidence for intersexual chemical mimicry in a dioecious plant. Ecol Lett. 2012;15:978–85. 42. Darwell CT. Al-Beidh S, cook JM. Molecular species delimitation of a symbiotic fig-pollinating wasp species complex reveals extreme deviation from reciprocal partner specificity. BMC Evol Biol. 2014;14:189. 43. Grandi G. Hyménoptères sycophiles récoltés aux iles Philippines par C.F. Baker, I. Agaonini. Philipp J Sci. 1927;33:309–29. 44. Darwell CT, Cook JM. Cryptic diversity in a fig wasp community – morphologically differentiated species are sympatric but cryptic species are allopatric. Mol Ecol. 2017;26:937–50. 45. Bain A, Tzeng HY, Wu WJ, Chou LS. Ficus (Moraceae) and fig wasps (Hymenoptera: Chalcidoidea) in Taiwan. Bot Stud. 2015;56:11. 46. Sutton TL, DeGabriel JL, Riegler M, Cook JM. Local coexistence and genetic isolation of three pollinator species on the same fig tree species. Heredity. 2017; doi:10.1038/hdy.2016.125. 47. Kerdelhué C, Hochberg ME, Rasplus JY. Active pollination of Ficus sur by two sympatric fig wasp species in West Africa. Biotropica. 1997;29:69–75. 48. Tian E, Nason JD, Machado C, Zheng L, Yu H, Kjellberg F. Lack of genetic isolation by distance, similar genetic structuring but different demographic histories in a fig-pollinating wasp mutualism. Mol Ecol. 2015;24:5976–91. 49. Liu M, Zhang JI, Chen Y, Compton SG, Chen XY. Contrasting genetic responses to population fragmentation in a coevolving fig and fig wasp across a mainland–island archipelago. Mol Ecol. 2013;22:4384–96. 50. Liu M, Rui Zhao R, Chen Y, Zhang J, Compton SG, Chen XY. Competitive exclusion among fig wasps achieved via entrainment of host plant flowering phenology. PLoS One. 2014;9:e97783. doi:10.1371/journal.pone.0097783. 51. Kjellberg F, Proffit M. Tracking the elusive history of diversification in plant- herbivorous insect-parasitoid food webs: insights from figs and fig-wasps. Mol Ecol. 2016;25:843–5. 52. Sutton TL, Riegler M, Cook JM. One step ahead: a parasitoid disperses farther and forms a wider geographic population than its fig wasp host. Mol Ecol. 2016;25:882–94. 53. Brown RM, Siler CD, Oliveros CH, Esselstyn JA, Diesmos AC, Hosner PA, et al. Evolutionary processes of diversification in a model island archipelago. Annu Rev Ecol Evol Syst. 2013;44:411–35. 54. Esselstyn JA, Maher SP, Brown RM. Species interactions during diversification and community assembly in an island radiation of shrews. PLoS One. 2011; 6:e21885. 55. Azuma H, Harrison RD, Nakamura K, Su ZH. Molecular phylogenies of figs and fig-pollinating wasps in the Ryukyu and Bonin (Ogasawara) islands. Japan Genes Genet Syst. 2010;85:177–92. 56. Lin RC, Yeng CKL, Li SH. Drastic post-LGM expansion and lack of historical genetic structure of a subtropical fig-pollinating wasp (Ceratosolen sp. 1) of Ficus septica in Taiwan. Mol Ecol. 2008;17:5008–22. 57. Machado CA, Jousselin E, Kjellberg F, Compton SG, Herre EA. Phylogenetic relationships, historical biogeography and character evolution of fig- pollinating wasps. Proc R Soc Lond B. 2001;268:685–94. Submit your next manuscript to BioMed Central and we will help you at every step:

• We accept pre-submission inquiries • Our selector tool helps you to find the most relevant journal • We provide round the clock customer support • Convenient online submission • Thorough peer review • Inclusion in PubMed and all major indexing services • Maximum visibility for your research

Submit your manuscript at www.biomedcentral.com/submit